[hts-users:03663] Re: Re synthesis with Straight voice build

while converting utts files to lab files using make command we got following errors:

nilesh-pc@nilesh-pc-desktop:~/Desktop/TTS_HTS_Workshop/DAIICTT_GUJ_HTS$ make
make[1]: Entering directory `/home/nilesh-pc/Desktop/TTS_HTS_Workshop/DAIICTT_GUJ_HTS/data'
# Extracting monophone and fullcontext labels from utterance files
mkdir -p labels/mono
mkdir -p labels/full
for utt in utts/*.utt; do \
        base=`basename ${utt} .utt`; \
        if [ -s ${utt} ]; then \
            /home/nilesh-pc/speech/installations/festival/examples//dumpfeats    -eval     scripts/extra_feats.scm \
                    -relation Segment \
                    -feats    scripts/label.feats \
                    -output   tmp \
                    ${utt}; \
        fi; \
        gawk -f scripts/label-full.awk tmp > labels/full/${base}.lab; \
        gawk -f scripts/label-mono.awk tmp > labels/mono/${base}.lab; \
    done
rm -f tmp
# Generating monophone and fullcontext Master Label Files (MLF)
echo "#!MLF!#" > labels/mono.mlf
echo "\"*/*.lab\" -> \"/home/nilesh-pc/Desktop/TTS_HTS_Workshop/DAIICTT_GUJ_HTS/data/labels/mono\"" >> labels/mono.mlf
echo "#!MLF!#" > labels/full.mlf
echo "\"*/*.lab\" -> \"/home/nilesh-pc/Desktop/TTS_HTS_Workshop/DAIICTT_GUJ_HTS/data/labels/full\"" >> labels/full.mlf
# Generating a fullcontext model list file
mkdir -p lists
rm -f tmp
for lab in labels/full/*.lab; do \
        if [ -s ${lab} -a -s labels/mono/`basename ${lab}` -a -s cmp/`basename ${lab} .lab`.cmp ]; then \
            sed -e "s/.* //g" ${lab} >> tmp; \
        fi \
    done
sort -u tmp > lists/full.list
sort: open failed: tmp: No such file or directory
make[1]: *** [list] Error 2
make[1]: Leaving directory `/home/nilesh-pc/Desktop/TTS_HTS_Workshop/DAIICTT_GUJ_HTS/data'
make: *** [data] Error 2

Can you please help me how to solve this error.

On Wed, Mar 6, 2013 at 1:13 PM, Csapo Tamas Gabor <csapot@xxxxxxxxxxx> wrote:

Hi Sarah,

1) you need to convert utt files to lab files
-> you can find an example for this in data/Makefile, 'labels' part
2) copy your new lab files into data/labels/gen
3) include them in the training:
-> in the data directory, 'make list' and 'make scp'
4) modify your scripts/Config.pm and include only the necessary switches:
-> e.g. MKUNG, MKUN1, PGEN1, WGEN1

Hope this helps.

Best,
Tamas

2013.03.06. 5:33 keltezéssel, Sarah Jones írta:

Hi,

*.utt files can be generated by festival speech synthesis system for the
unseen sentences.

As said in my last mail, I already have .utt files generated from
festival. What I want to know is, how to use the models saved in
models/qst001/ver001 directory and generate parameter sequences for
respective test .utt sentences, and use such parameters as input to the
matlab-straight code for waveform generation.

Thank you,
- Sarah

On Mon, Mar 4, 2013 at 9:32 PM, Keiichiro Oura <uratec@xxxxxxxxxxxxxxx
<mailto:uratec@xxxxxxxxxxxxxxx>> wrote:

Hi,

All model files are in models/qst001/ver001 directory.
*.utt files can be generated by festival speech synthesis system for
the unseen sentences.

Regards,
Keiichiro Oura

2013/3/2 Sarah Jones <jonessarah346@xxxxxxxxx
<mailto:jonessarah346@gmail.com>>:

> Hi All,
>
> I have build voices using demo scripts for both STRAIGHT and
non-STRAIGHT
> versions from HTS download page. I found that non-straight
version creates
> model files in voices/qst001/ver1/ folder. But the straight
version doesn't
> create, probably because straight is not supported by
hts_engine_api. In
> this regard, I have few questions:
>
> 1. Where are the model files located in straight voice build.
>
> 2. How to synthesize speech using straight based models for an
unknown
> English sentence, assuming I have .utt file. What are the sequence of
> commands involved in the process.
>
> Thank you All,
> - Sarah